尊敬的用户您好，这是来自FT中文网的温馨提示：如您对更多FT中文网的内容感兴趣，请在苹果应用商店或谷歌应用市场搜索“FT中文网”，下载FT中文网的官方应用。
As recent FT analysis highlights, corporate America may be waxing lyrical about the promise of artificial intelligence but few boardrooms appear able to describe how the technology is actually changing their businesses for the better. There is, however, one sector where the gains are clear, even if it is less eye-catching to profit-chasing investors: public health. For an industry with intensely high demands on accuracy and efficiency, generative AI could transform healthcare delivery and patient outcomes. In turn, the potential benefits for society, the economy and stretched public budgets are immense.
正如英国《金融时报》最近的分析所强调的那样，尽管美国企业界对人工智能的前景赞不绝口，但很少有董事会能说明这种科技究竟如何实实在在地改善了他们的业务。不过，有一个领域的收益已相当明确，尽管它对逐利的投资者而言不那么吸睛：公共卫生。对于一个对准确性和效率要求极高的行业，生成式AI有望重塑医疗服务的提供方式并改善患者预后。由此给社会、经济以及紧张的公共预算方面带来的潜在收益都将十分可观。
The greatest payback from AI may well come from the earlier and more accurate detection of life-threatening illnesses. In June, Microsoft claimed it had built a diagnostic medical tool that was four times more successful than doctors at determining complex ailments. Some models may even be powerful enough to ascertain distant health risks. Last month, scientists using the gen-AI system Delphi-2M, which was built at the European Molecular Biology Laboratory in Cambridge and trained on large-scale health records, reported that it could predict susceptibility to more than 1,000 diseases decades into the future.
AI带来的最大回报，很可能来自对威胁生命的疾病更早地作出更准确的检测。6月，微软(Microsoft)宣称，其打造的一款医学诊断工具在判定复杂病症方面的成功率是医生的四倍。有些模型甚至或许强大到能够判断远期健康风险。上个月，使用剑桥(Cambridge)的欧洲分子生物学实验室(EMBL)构建、并以大规模健康记录进行训练的生成式AI系统Delphi-2M的科研人员报告称，该系统能够预测未来数十年内对一千多种疾病的易感性。
But AI’s impact extends well beyond preventive support. In hospitals, the technology can rapidly analyse X-rays, CAT scans and MRIs. Robotic surgery systems powered by AI can improve surgical precision. Labs are harnessing large language models to accelerate drug discovery too. Crucially, all these applications complement health professionals and free them to provide better care to more patients.
但AI的影响远超预防性支持。在医院，这项科技可快速分析X光、CAT扫描和MRI。由AI驱动的机器人手术系统可提升手术精度。实验室也在利用大型语言模型加速药物研发。关键在于，这些应用补充了医护专业人员的工作，让他们得以为更多患者提供更优质的护理。
Less glamorous but equally significant is AI’s ability to cut administrative burdens. The US-based Commonwealth Fund estimates that paperwork costs, linked in part to onerous insurance checks, could account for about 30 per cent of America’s excess per capita health spending compared with other nations. In surgeries and hospitals, speech-processing technologies can also be used to transcribe conversations with patients, create structured medical notes and draft letters. A recent study by London’s Great Ormond Street Hospital found a more than 50 per cent reduction in documentation time for clinicians using so-called ambient voice technologies.
AI削减行政负担的能力或许不够亮眼，却同样意义重大。总部位于美国的基金会Commonwealth Fund估计，文书工作成本（部分源于繁琐的保险核查）可能占美国人均医疗支出超出其他国家部分的约30%。在手术室和医院，语音处理技术还能转录医患对话、生成结构化病历并起草信函。伦敦大奥蒙德街儿童医院(Great Ormond Street Hospital)的最新研究显示，采用环境语音技术后，临床医生的文书处理时间减少了50%以上。
Clearly, accelerating AI adoption should be a priority for governments worldwide. Ageing populations and the growing prevalence of chronic diseases in advanced economies are contributing to rising healthcare costs. This strains already stretched public budgets and makes it harder for individuals to afford private cover, as the ongoing wrangling between Republicans and Democrats over insurance support reflects. The World Health Organization also projects a shortage of around 11mn healthcare workers by 2030, which will be more pronounced in lower and middle-income countries. As the Covid-19 pandemic demonstrated, a healthier international population helps reduce the spread of disease, benefiting everyone.
显然，加速采用AI应当成为全球各国政府的优先事项。发达经济体的人口老龄化和慢性疾病日益普遍正推高医疗成本，这不仅进一步拉紧本已捉襟见肘的公共预算，也让个人更难负担私人保险——共和党与民主党就保险补助的长期争执正反映了这一点。世界卫生组织(World Health Organization)还预测，到2030年全球将短缺约1100万名医护人员，这一缺口在中低收入国家将更为突出。正如新冠肺炎大流行所示，更健康的全球人口有助于减少疾病传播，最终惠及所有人。
For all the upsides, the use of AI in healthcare is still nascent and patchy. This is partly because the technology is still developing, and rigorous testing is needed before its widespread use in medicine. Health professionals need to train with it too. Data privacy concerns, fragmented sharing networks and outdated IT systems add further complications. Managers can also be reticent over introducing AI where staff may feel their jobs are under threat, or in private systems where revenue depends on high service volumes.
尽管有诸多优势，AI在医疗领域的应用仍处于早期且分布零散。这部分是因为科技尚在发展阶段，在广泛用于医疗之前需要经过严格测试。医护专业人员也需要接受相关培训。数据隐私顾虑、碎片化的共享网络以及陈旧的IT系统进一步增加了复杂性。管理者也可能对引入AI持谨慎态度，尤其是在员工可能感觉岗位受威胁的情况下，或在依赖高服务量获取收入的私营体系中。
A concerted push from governments, health regulators and tech companies is needed to help fund, trial and deploy AI applications in hospitals, and overcome the cultural and technical implementation obstacles. This will not be easy, but the scale of the prize — in the form of healthier populations and more sustainable healthcare systems — ought to focus the minds.
政府、卫生监管机构和科技公司需要协同发力，支持在医院内为AI应用提供资金、开展试验并实现部署，同时克服文化和技术层面的落地障碍。此事并不容易，但其回报之大——更健康的人口与更可持续的医疗体系——理应引起高度重视。